Skip to main content

Abstraction is not a principle of Object Oriented Programming

TL;DR

Abstraction is a core concept in programming and not a principle that is solely applicable in OOP, because programming is all about dealing with an abstract representation of the business model (application space).

Abstraction is an overall programming principle

There are a lot of articles saying that OOP has four principles:

  • Encapsulation
  • Inheritance
  • Polymorphism
  • Abstraction

But if we do a web searh for what is abstraction? we get more or less the same definition as the one from here.

Abstraction (from the Latin abs, meaning away from and trahere, meaning to draw) is the process of taking away or removing characteristics from something in order to reduce it to a set of essential characteristics.

Although the article defines abstraction in an OOP context, abstraction is actually a an overall programming principle.

Why? Because programming means modeling entities and interactions of an abstract representation of the business model. And since there are several programming paradigms all of which are used to represent a set of specific business models all of those paradigms use abstraction to succeed in that representation.

An example of abstraction in OOP

To better clarify what abstraction is (in OOP) let's consider a trivial example: A StudentRegistrationController (ASP.NET MVC) needs to save the data from a registration form (Student) into some sort of database represented by an instance of IStudentDataStore.

For the sake of brevity let's say that IStudentDataStore interface has the following definition:

  public interface IStudentDataStore
  {
      void Save(Student student);
  }

With that in mind, the code for the example above would look like this:

  public class Student
  {
      public string FirstName { get; set; }

      public string LastName { get; set; }

      public DateTime DateOfBirth { get; set; }

      public Gender Gender { get; set; }

      public string Email { get; set; }

      public string PhoneNumber { get; set; }
  }

  public class StudentRegistrationController : Controller
  {
      public StudentRegistrationController(IStudentDataStore dataStore)
      {
	  _dataStore = dataStore;
      }

      public ActionResult Save(Student viewModel)
      {
	  // Validation ommited for brevity
	  _dataStore.Save(viewModel);
      }
  }

As we can see from the code the IStudentDataStore abstracts the details of how Student data is persisted. The interface doesn't tell whether the data is persisted to a flat file, SQL Database, NoSQL database or other media. As long as we have an implementation of IStudentDataStore the controller works just fine.

But is this abstraction technique available only in OOP paradigm? The answer is no. The same thing, although in a different form can be used in other paradigms. To demonstrate so, let's use lisp as a programming language and let's see how abstraction works in a functional programming paradigm.

The same example of abstraction in functional programming

To see how the same thing would be written in a functional style let's first see what are the entities from the previous example and how those entities interact between them:

  • StudentRegistrationController
    • Is the top-most entity
    • Has an IStudentDataStore on which calls the Save method
  • IStudentDataStore
    • Defines a contract between the controller and the underlying data store
  • Student
    • Is a data contract, i.e. encapsulates all the properties of a student into a single entity

Having the above and knowing that in functional programming everything is a function we replace every entity with a function:

  • We'll create a top-level function named save-student; it will have the following parameters:
    • student-info -> a hash map containing the data contract
    • persist-func -> a function that will be called to save the data to a data store
  • Another two functions persist-student-info-to-file and persist-student-info-in-memory will encapsulate the logic of persisting data to a data store.

Note that both functions have the same signature. The parameters of these functions represent the data contract.

  (defun persist-student-info-to-file (first-name last-name dob gender email phone)
    ;; Writes the info as a new line into the file specified by *database-location*
    (let ((stream (open *database-location*
		:direction :output
		:if-exists :append)))
      (format stream "~s|~s|~s|~s|~s|~s"
	  first-name
	  last-name
	  dob
	  gender
	  email
	  phone)
      (close stream)))


  (defun persist-student-info-in-memory (first-name last-name dob gender email phone)
    ;; Persists student data into an in-memory data store named *all-students*
    (let ((student-id (list-length *all-students*)))
      (setq *all-students*
	(append *all-students*
	    (list student-id first-name last-name dob gender email phone)))))

  (defun save-student (student-info persist-func)
    ;; Saves student data to a persistent store
    (let ((first-name (gethash 'first-name student-info))
	  (last-name (gethash 'last-name student-info))
	  (dob (gethash 'date-of-birth student-info))
	  (gender (gethash 'gender student-info))
	  (email (gethash 'email student-info))
	  (phone (gethash 'phone-number student-info)))
      (funcall persist-func first-name last-name dob gender email phone)))

Having the definitions above we can achieve the same level of abstraction by putting everything together and defining global variables (think of it as poor mans dependency injection):

  (defparameter *database-location* "/tmp/students")
  (defparameter *student-info* (make-hash-table))

  (progn
    (setf (gethash 'first-name *student-info*) "John")
    (setf (gethash 'last-name *student-info*) "Doe")
    (setf (gethash 'date-of-birth *student-info) "2000-05-14")
    (setf (gethash 'email *student-info*) "john.doe@example.com")
    (setf (gethash 'phone-number *student-info*) "1234567981"))

  (save-student *student-info* 'persist-student-info-to-file)

Et voilà! We have used abstraction in a functional style to remove the details of how the data is persisted thus showing that abstraction isn't a principle applicable only to object oriented programming.

Acknowledgments

A big thank you to Ion Cojocaru and Florin Olariu who reviewed this post before publication.

Comments

Comments powered by Disqus